Sources and methodology
Every rate on this site comes from one dated snapshot of New York Fed data, currently through October 8, 2026. This page lists the endpoints, the formulas we apply to them and the checks we run before publishing.
Data sources
Five series are downloaded from the public New York Fed Markets API. Each request asks for every observation from April 2, 2018, the first SOFR value date, to the download date. The API returns the published rate, the 1st, 25th, 75th and 99th volume-weighted percentiles, volume in billions and a revision indicator; the EFFR records also carry the federal funds target range in force, and the SOFR Averages and Index records carry the three averages and the index level.
| Series | API request | First date | Last date | Observations |
|---|---|---|---|---|
| SOFR | /api/rates/secured/sofr/search.json | Apr 2, 2018 | Oct 8, 2026 | 2,129 |
| SOFR Averages and Index (SOFRAI) | /api/rates/secured/sofrai/search.json | Mar 2, 2020 | Oct 9, 2026 | 1,652 |
| TGCR | /api/rates/secured/tgcr/search.json | Apr 2, 2018 | Oct 8, 2026 | 2,129 |
| BGCR | /api/rates/secured/bgcr/search.json | Apr 2, 2018 | Oct 8, 2026 | 2,129 |
| EFFR | /api/rates/unsecured/effr/search.json | Apr 2, 2018 | Oct 8, 2026 | 2,141 |
All requests go to https://markets.newyorkfed.org with startDate and endDate parameters. The SOFR Averages and Index start on March 2, 2020 because that is when the New York Fed began publishing them; the indicative history it posts for earlier dates is not official and is not used here. In the October 10, 2026 download, EFFR had 12 more observations than SOFR because it was published on dates when the repo rates were not, such as April 3, 2026 and July 3, 2026. Methodology for each rate is the New York Fed's own, described on its reference-rate methodology page and summarized in how SOFR is calculated.
Snapshot date and update practice
The snapshot is refreshed by hand. A person runs a short script that downloads the five series, checks that each response holds more than 1,000 records, and rebuilds the site. Visitors' browsers never call the New York Fed; the calculators read copies of the same files served from this domain.
The current snapshot runs through SOFR for October 8, 2026 and SOFR Averages and Index for October 9, 2026. Because the New York Fed publishes SOFR one business day after its value date, the averages and index carry a date one business day later than the last SOFR print they include. Every data page and calculator states the "data through" date under its title, and the rate strip at the top of each page is labeled with the SOFR value date; the 30- and 90-day averages in it carry the later SOFR Averages date.
The New York Fed may revise SOFR, TGCR and BGCR at about 2:30 p.m. ET on the day of first publication, and only if the change exceeds one basis point. A snapshot taken after that window holds the final figure; one taken the same morning as a print could miss a revision. The October 10, 2026 download, for example, came the day after the October 8 print was published, once its revision window had closed.
How derived statistics are computed
Everything below is our arithmetic on the New York Fed data, not a figure the New York Fed publishes. It is recomputed from the raw files at every build.
Annual statistics
For each calendar year we report the number of published SOFR observations, the first and last print, the simple mean and median of the daily prints, the high and low with the first date each occurred, mean and maximum daily volume, and the five largest moves between consecutive prints within the calendar year. The first print of a year is not compared with the last print of the year before, so a turn-of-year move is left out of both years: the −12 bp move from December 31, 2025 to January 2, 2026, for example, does not appear in the 2026 list. The mean is unweighted: a Friday print counts once, even though it accrues for three calendar days in a loan. For a calendar-weighted figure use the compounded SOFR calculator.
Percentile rank
The rank of the latest SOFR is the share of all observations since April 2, 2018 that are at or below it. On October 8, 2026 that was 65 percent.
Month-end effect
For each month with at least six SOFR observations, the month-end effect is the last published print of the month minus the simple mean of the five prints before it, in basis points. The measure makes no adjustment for policy moves. The largest reading, +59.2 bp for July 29, 2022, mostly reflects the target range rising from 1.50%–1.75% to 2.25%–2.50% effective July 28, 2022, not balance-sheet pressure. SOFR spikes and month-end separates the two.
Spreads and the target range
SOFR minus EFFR and SOFR minus TGCR are computed only on dates when both rates were published, as whole basis points per day, then summarized by year as mean, minimum and maximum. The federal funds target range history comes from the target fields on the EFFR records; a change is dated to the first EFFR value date on which the new range appears.
The compounding engine
All calculators share one routine for compounded SOFR in arrears. It follows the New York Fed and ARRC convention: daily compounding on business days, simple interest at the preceding business day's rate across weekends and holidays, actual days over a 360-day year.
Lookback, observation shift and lockout variants change which print is applied to each day or which dates bound the observation period; the arithmetic is the same. Custom periods measured off the published index use (Index_end / Index_start − 1) × 360 / days.
Verification against published values
We test the engine against the New York Fed's own outputs. Results against the October 10, 2026 snapshot:
- Compounding every SOFR print from April 2, 2018 reproduces all 1,652 published SOFR Index values to within 0.000000005, half a unit in the eighth decimal place where the index is rounded.
- Recomputing the 30-, 90- and 180-day SOFR Averages for every publication date, with the window starting exactly 30, 90 or 180 calendar days earlier, matches all 1,652 published 30-day values once rounded to five decimals. The largest unrounded difference for any of the three averages was 0.000005 percentage points.
- For periods of 21 business days or longer between published index dates, the engine and the index ratio differ by at most 0.000011 percentage points; for ten business days, by at most 0.000023.
Over a single day the index ratio can differ from the daily print by up to about 0.00033 percentage points. That is the index's eight-decimal rounding divided across one day of accrual, not an error in either method.
Day-count conventions
The day count calculator and the interest tools implement ACT/360, ACT/365 Fixed, 30/360 US (bond basis, 2006 ISDA Definitions section 4.16(f), without the February end-of-month adjustment), 30E/360 (section 4.16(g)) and ACT/ACT ISDA, which splits a period at each January 1 and divides by 365 or 366. The start date accrues and the end date does not.
ISDA spread adjustments
The USD LIBOR fallback spreads used by the LIBOR fallback calculator are the seven values fixed on March 5, 2021, taken from Bloomberg Index Services Limited's technical notice of that date. Five of them are also codified in the Federal Reserve's Regulation ZZ at 12 CFR 253.4(c), and we checked that the values agree.
| USD LIBOR tenor | Spread adjustment (%) | Spread adjustment (bp) |
|---|---|---|
| Overnight | 0.00644 | 0.644 |
| 1 week | 0.03839 | 3.839 |
| 1 month | 0.11448 | 11.448 |
| 2 months | 0.18456 | 18.456 |
| 3 months | 0.26161 | 26.161 |
| 6 months | 0.42826 | 42.826 |
| 12 months | 0.71513 | 71.513 |
Bloomberg's notice states: "The Data, including any sample calculations, are for illustrative purposes only. Neither Bloomberg nor ISDA guarantees the timeliness, accurateness, completeness of, or fitness for a particular purpose with respect to, the Data and each shall have no liability in connection with the Data."
What this site does not do
- No Term SOFR values. CME Term SOFR is a licensed CME Group benchmark. We explain it in Term SOFR vs overnight SOFR but never display, estimate or calculate with its rates.
- No forecasts. Nothing here projects SOFR or the target range. The rate shock calculator applies the moves you choose; it does not predict any.
- No live feed. Figures are as of the snapshot date shown on each page, never streamed.
- No contract terms. Calculators follow standard conventions. Your agreement may specify a different lookback, floor, rounding or day count, and the agreement governs.
New York Fed terms of use
The New York Fed licenses its reference-rate data for reuse under its Terms of Use, which ask anyone republishing the data to carry a notice. Ours, completed in the form the Terms set out:
The SOFR, SOFR Averages, SOFR Index, TGCR, BGCR and EFFR data are subject to the Terms of Use posted at newyorkfed.org. The New York Fed is not responsible for publication of the SOFR, SOFR Averages, SOFR Index, TGCR, BGCR and EFFR data by Right Angles Technologies Inc (BitsSecurity), does not sanction or endorse any particular republication, and has no liability for your use.
Because rate names appear in our calculator names, the Terms also require this statement:
BitsSecurity (Right Angles Technologies Inc) is not affiliated with the New York Fed. The New York Fed does not sanction, endorse, or recommend any products or services offered by BitsSecurity (Right Angles Technologies Inc).
The Terms reproduce a notice about data licensed from DTCC Solutions LLC, which the New York Fed used before January 24, 2022. We repeat it because our history covers that period: "The Secured Overnight Financing Rate (SOFR) Data and Broad General Collateral Rate (BGCR) Data are calculated using data provided under a license granted to the New York Fed by DTCC Solutions LLC ("Solutions"), an affiliate of The Depository Trust & Clearing Corporation. Solutions, its affiliates, and third parties from which they obtained data have no liability for the content of this material."
Averages we compute ourselves, yearly statistics, spreads, percentile ranks and charts are modifications of New York Fed content made by BitsSecurity and should not be attributed to the New York Fed.
Corrections
If a figure here disagrees with the New York Fed's published value for the same date, the New York Fed's figure is correct. Email info@rightangles.com with the series, the date and both values; when we confirm an error we fix the source data or formula and rebuild every affected page. Questions about how the New York Fed calculated a given day's rate belong with the administrator, which takes them in writing at rateproduction@ny.frb.org. Other ways to reach us are on the contact page.